Book Image

Hands-On Financial Trading with Python

By : Jiri Pik, Sourav Ghosh
Book Image

Hands-On Financial Trading with Python

By: Jiri Pik, Sourav Ghosh

Overview of this book

Creating an effective system to automate your trading can help you achieve two of every trader’s key goals; saving time and making money. But to devise a system that will work for you, you need guidance to show you the ropes around building a system and monitoring its performance. This is where Hands-on Financial Trading with Python can give you the advantage. This practical Python book will introduce you to Python and tell you exactly why it’s the best platform for developing trading strategies. You’ll then cover quantitative analysis using Python, and learn how to build algorithmic trading strategies with Zipline using various market data sources. Using Zipline as the backtesting library allows access to complimentary US historical daily market data until 2018. As you advance, you will gain an in-depth understanding of Python libraries such as NumPy and pandas for analyzing financial datasets, and explore Matplotlib, statsmodels, and scikit-learn libraries for advanced analytics. As you progress, you’ll pick up lots of skills like time series forecasting, covering pmdarima and Facebook Prophet. By the end of this trading book, you will be able to build predictive trading signals, adopt basic and advanced algorithmic trading strategies, and perform portfolio optimization to help you get —and stay—ahead of the markets.
Table of Contents (15 chapters)
1
Section 1: Introduction to Algorithmic Trading
3
Section 2: In-Depth Look at Python Libraries for the Analysis of Financial Datasets
9
Section 3: Algorithmic Trading in Python

Introduction to statsmodels

statsmodels is a Python library that allows us to explore data, perform statistical tests, and estimate statistical models.

This chapter focuses on statsmodels' modeling, analysis, and forecasting of time series.

Normal distribution test with Q-Q plots

An underlying assumption of many statistical learning techniques is that the observations/fields are normally distributed.

While there are many robust statistical tests for normal distributions, an intuitive visual method is known as a quantile-quantile plot (Q-Q plot). If a sample is normally distributed, its Q-Q plot is a straight line.

In the following code block, the statsmodels.graphics.api.qqplot(...) method is used to check if a numpy.random.uniform(...) distribution is normally distributed:

from statsmodels.graphics.api import qqplot
import numpy as np
fig = qqplot(np.random.uniform(size=10000), line='s')
fig.set_size_inches(12, 6)

The resulting plot depicted in...